Added 60dB integration

CustomSettings.py

@@ -8,6 +8,11 @@
 assistantSpeechOn = True
 offlineTTS = False # Set to true if you want to use an offline TTS instead of ElevenLabs or Google TTS
 elevenLabs = True # For a highly realistic English TTS
+sixtyDB = True # 60db.ai English TTS (used together with ElevenLabs)
+
+# English TTS provider priority: "elevenlabs" or "60db".
+# The chosen provider is used first; if it fails, the other is tried automatically, then offline TTS.
+ttsProvider = "elevenlabs"
 googleSTT = True # If this is False, a pretty bad STT in english will be used
 
 # Google Cloud Speech To Text and Text To Speech
@@ -74,6 +79,14 @@
 #VOICE_ID = "TxGEqnHWrfWFTfGW9XjX" # Josh (Snarky voice)
 
 
+# 60db.ai (English TTS) - get voice IDs from "GET https://api.60db.ai/myvoices"
+SIXTYDB_VOICE_ID = "{VOICE_ID_HERE}" # Leave as-is to use the 60db system default voice
+SIXTYDB_STABILITY = 30   # 0-100, lower = less stable & funnier (mirrors ElevenLabs STABILITY x100)
+SIXTYDB_SIMILARITY = 90  # 0-100, source voice matching (mirrors ElevenLabs SIMILARITY_BOOST x100)
+SIXTYDB_SPEED = 1.0      # 0.5-2.0 speech speed multiplier
+SIXTYDB_ENHANCE = True   # Audio quality improvement
+
+
 
 # Print out more outputs, for debugging (Developer Mode)
 devMode = False

README.md

@@ -1,5 +1,5 @@
 # SmartVoiceAssistant
-A smart AI voice assistant with multi-language support and long-term memory. Currently best for Swedish and English. Compatible with Windows and Raspberry Pi. The assistant can use various functions and tools to answer question (Google, Wolfram Alpha, etc.). Based on OpenAI's GPT-models, Google STT and TTS, and ElevenLabs TTS.
+A smart AI voice assistant with multi-language support and long-term memory. Currently best for Swedish and English. Compatible with Windows and Raspberry Pi. The assistant can use various functions and tools to answer question (Google, Wolfram Alpha, etc.). Based on OpenAI's GPT-models, Google STT and TTS, ElevenLabs TTS, and 60db.ai TTS.
 
 <img src="https://github.com/ottobjorkland/SmartVoiceAssistant/assets/81506445/015800af-8b39-46a6-be58-94af17b0179a" height="400">
 
@@ -13,6 +13,7 @@ If you do not want to gather all of this information or do not have time, simply
 - [OpenAI API Key](https://platform.openai.com/account/api-keys): If you have an OpenAI account, you can find your API Key in the user settings.
 - [Porcupine Access Key (Wake-word recognizer)](https://console.picovoice.ai/): Sign up for Picovoice Console to get your Access Key.
 - [ElevenLabs API Key (Text-To-Speech)](https://elevenlabs.io/): Create an account, click the profile icon in the top-right corner, and get the API Key from "Profile Settings".
+- [60db.ai API Key (Text-To-Speech)](https://60db.ai/): Create an account and get your API Key from the dashboard. Used as an alternative/fallback English TTS alongside ElevenLabs (see the "English Text-To-Speech providers" section below).
 - [Wolfram Alpha App ID](https://developer.wolframalpha.com/): Sign up for a developer account, create an app under "My Apps" > "Get an AppID", and get the AppID
 - Google
   - [Custom Search API Key (developerKey)](https://developers.google.com/custom-search/v1/overview): Create a new project and get the API Key by clicking "Get a Key".
@@ -26,6 +27,33 @@ If you do not want to gather all of this information or do not have time, simply
     1. A JSON key file will be downloaded to your device
     1. Move the JSON file to the repository folder that all other files Python and JSON files are in
 
+## English Text-To-Speech providers (ElevenLabs + 60db.ai)
+For English, the assistant supports two high-quality TTS providers: **ElevenLabs** and **60db.ai**. You pick one as the primary; if it fails (e.g. a network error or wrong API Key), the other is tried automatically, and if both fail it falls back to the offline TTS. Swedish always uses Google Cloud TTS.
+
+### Configure it in `CustomSettings.py`
+```python
+elevenLabs = True            # Enable ElevenLabs
+sixtyDB = True               # Enable 60db.ai
+ttsProvider = "elevenlabs"   # Which one to use first: "elevenlabs" or "60db" (the other is the automatic fallback)
+
+# 60db.ai voice & settings (English) - get voice IDs from "GET https://api.60db.ai/myvoices"
+SIXTYDB_VOICE_ID = "{VOICE_ID_HERE}"  # Leave as-is to use the 60db system default voice
+SIXTYDB_STABILITY = 30                # 0-100, lower = less stable & funnier
+SIXTYDB_SIMILARITY = 90               # 0-100, source voice matching
+SIXTYDB_SPEED = 1.0                   # 0.5-2.0 speech speed multiplier
+SIXTYDB_ENHANCE = True                # Audio quality improvement
+```
+Add your 60db API Key in `apiKeys.py`:
+```python
+SIXTYDB_API_KEY = "{KEY_HERE}"
+```
+
+### Notes
+- To make 60db the primary voice, set `ttsProvider = "60db"`.
+- To disable a provider entirely, set its flag (`elevenLabs` / `sixtyDB`) to `False`.
+- Finding your 60db voice IDs: send a `GET` request to `https://api.60db.ai/myvoices` with the header `Authorization: Bearer YOUR_API_KEY`; copy a `voice_id` from the response into `SIXTYDB_VOICE_ID`.
+- No extra packages are required for 60db (it uses `requests` and the built-in `base64`, which are already installed).
+
 ## Raspberry Pi
 ### Set-Up
 **Run these commands to install packages on a Raspberry Pi (terminal):**

VoiceAssistant_5.3.py

@@ -1,4 +1,4 @@
-import os, time, math, re, json, pyttsx3, requests, pvporcupine, struct, pyaudio, datetime, tiktoken
+import os, time, math, re, json, base64, pyttsx3, requests, pvporcupine, struct, pyaudio, datetime, tiktoken
 import speech_recognition as sr
 from colorama import init, Fore, Back, Style
 from urllib.request import urlopen
@@ -42,7 +42,9 @@
 from CustomSettings import wakeUpWords, STABILITY, SIMILARITY_BOOST, VOICE_ID, animationFPS, googleSTT, swedish, english, maxToolsPerPrompt, openAIdelay
 from CustomSettings import googleTTS_name, googleTTS_gender, swedishStartPrompt, wakeWordOn, wakeSpeaker, speakerSleepTime, RaspberryPi, devMode, overrideMemPrompt
 from CustomSettings import sweOverrideMemPrompt, GPT4, elevenLabs, wolframAlpha, googleSearch
+from CustomSettings import sixtyDB, ttsProvider, SIXTYDB_VOICE_ID, SIXTYDB_STABILITY, SIXTYDB_SIMILARITY, SIXTYDB_SPEED, SIXTYDB_ENHANCE
 from apiKeys import openai_api_key, porcupineAccessKey, XI_API_KEY, googleCustomSearchAPI, googleSearchEngineID, GOOGLE_JSON_CREDENTIALS, wolframAlphaAppID
+from apiKeys import SIXTYDB_API_KEY
 import openai
 openai.api_key = openai_api_key
 
@@ -719,79 +721,143 @@ def on_end_reached(event):
             pygame.init()
             pygame.mixer.init()
 
+def elevenLabsTTS(text):
+    # Generate English speech with ElevenLabs and write it to textToSpeechFilePath
+    headers = {
+        "Accept": "audio/mpeg",
+        "xi-api-key": XI_API_KEY,
+        "Content-Type": "application/json"
+    }
+    data = {
+        "text": text,
+        "voice_settings": {
+            "stability": STABILITY,
+            "similarity_boost": SIMILARITY_BOOST
+        }
+    }
+    if vlcLib == True: # For Raspberry Pi
+        response = requests.post(f"https://api.elevenlabs.io/v1/text-to-speech/{VOICE_ID}", json=data, headers=headers)
+        response.raise_for_status()
+        with open(textToSpeechFilePath, "wb") as f:
+            f.write(response.content)
+    else: # Streamed (for Windows/pygame and as the default)
+        response = requests.post(f"https://api.elevenlabs.io/v1/text-to-speech/{VOICE_ID}/stream", json=data, headers=headers, stream=True)
+        response.raise_for_status()
+        with open(textToSpeechFilePath, 'wb') as f:
+            for chunk in response.iter_content(chunk_size=1024):
+                if chunk:
+                    f.write(chunk)
+
+def sixtyDBtts(text):
+    # Generate English speech with 60db.ai and write it to textToSpeechFilePath.
+    # 60db returns JSON with the audio as a base64 string, so we decode it before writing.
+    headers = {
+        "Authorization": f"Bearer {SIXTYDB_API_KEY}",
+        "Content-Type": "application/json"
+    }
+    data = {
+        "text": text,
+        "stability": SIXTYDB_STABILITY,   # 0-100
+        "similarity": SIXTYDB_SIMILARITY, # 0-100
+        "speed": SIXTYDB_SPEED,           # 0.5-2.0
+        "enhance": SIXTYDB_ENHANCE,
+        "output_format": "mp3"            # keep mp3 so it plays from textToSpeech.mp3 like every other provider
+    }
+    if SIXTYDB_VOICE_ID and "{" not in SIXTYDB_VOICE_ID: # Otherwise use the 60db system default voice
+        data["voice_id"] = SIXTYDB_VOICE_ID
+    response = requests.post("https://api.60db.ai/tts-synthesize", json=data, headers=headers)
+    response.raise_for_status()
+    payload = response.json()
+    if not payload.get("success") or not payload.get("audio_base64"):
+        raise RuntimeError(payload.get("message", "60db returned no audio"))
+    with open(textToSpeechFilePath, "wb") as f:
+        f.write(base64.b64decode(payload["audio_base64"]))
+
+def englishTTSproviders():
+    # Ordered list of (name, generatorFunc) for English TTS.
+    # The provider chosen in ttsProvider comes first, the other is the automatic fallback.
+    allProviders = {
+        "ElevenLabs": (elevenLabs, elevenLabsTTS),
+        "60db": (sixtyDB, sixtyDBtts),
+    }
+    primary = "60db" if str(ttsProvider).lower() in ("60db", "sixtydb") else "ElevenLabs"
+    secondary = "ElevenLabs" if primary == "60db" else "60db"
+    order = []
+    for name in (primary, secondary):
+        enabled, generate = allProviders[name]
+        if enabled:
+            order.append((name, generate))
+    return order
+
+def disableTTSprovider(name):
+    # Disable a provider for the rest of the session (e.g. after a wrong API key) so we stop retrying it
+    global elevenLabs, sixtyDB
+    if name == "ElevenLabs": elevenLabs = False
+    elif name == "60db": sixtyDB = False
+    print(Style.BRIGHT+Fore.RED+f"WARNING: Disabling {name} for this session. Please provide a valid API Key in apiKeys.py to use {name}.")
+
+def isAuthError(e):
+    # True if the exception is an HTTP 401/403 (likely a wrong/missing API key)
+    response = getattr(e, "response", None)
+    return response is not None and getattr(response, "status_code", None) in (401, 403)
+
 def textToSpeech(text, language):
-
-    global elevenLabs
 
     print("Generating text-to-speech...")
 
-    if language == "sv": # Swedish
-        client = texttospeech.TextToSpeechClient()
-        input_text = texttospeech.SynthesisInput(text=text)
+    if language == "sv": # Swedish (Google Cloud TTS)
+        try:
+            client = texttospeech.TextToSpeechClient()
+            input_text = texttospeech.SynthesisInput(text=text)
 
-        if googleTTS_gender == "MALE": ssml_gender=texttospeech.SsmlVoiceGender.MALE
-        else: ssml_gender=texttospeech.SsmlVoiceGender.FEMALE
+            if googleTTS_gender == "MALE": ssml_gender=texttospeech.SsmlVoiceGender.MALE
+            else: ssml_gender=texttospeech.SsmlVoiceGender.FEMALE
 
-        voice = texttospeech.VoiceSelectionParams(
-            language_code="sv-SE",
-            name=googleTTS_name,
-            ssml_gender=ssml_gender,
-        )
-        audio_config = texttospeech.AudioConfig(
-            audio_encoding=texttospeech.AudioEncoding.MP3
-        )
-        response = client.synthesize_speech(
-            request={"input": input_text, "voice": voice, "audio_config": audio_config}
-        )
-        with open(textToSpeechFilePath, "wb") as out:
-            out.write(response.audio_content)
-
-    else: # English
-        if elevenLabs:
-            headers = {
-                "Accept": "audio/mpeg",
-                "xi-api-key": XI_API_KEY,
-                "Content-Type": "application/json"
-            }
-            data = {
-                "text": text,
-                "voice_settings": {
-                    "stability": STABILITY,
-                    "similarity_boost": SIMILARITY_BOOST
-                }
-            }
-            try:
-                if vlcLib == True: # For Raspberry Pi
-                    response = requests.post(f"https://api.elevenlabs.io/v1/text-to-speech/{VOICE_ID}", json=data, headers=headers)
-                    with open(textToSpeechFilePath, "wb") as f:
-                        f.write(response.content)
-                elif pygameLib == True: # For Windows
-                    response = requests.post(f"https://api.elevenlabs.io/v1/text-to-speech/{VOICE_ID}/stream", json=data, headers=headers, stream=True)
-                    with open(textToSpeechFilePath, 'wb') as f:
-                        for chunk in response.iter_content(chunk_size=1024):
-                            if chunk:
-                                f.write(chunk)
-            except Exception as e: # There was an error
-                print(Style.BRIGHT+Fore.RED+f"ElevenLabs Error:\n{e}")
-                print("Using offline text-to-speech instead...")
-                offlineTextToSpeech(text)
-                return None
-        else: # elevenLabs == False
-            offlineTextToSpeech(text)
-            return None
-
-    try:
-        playAudio(language)
-    except Exception as e:
-        print(Style.BRIGHT+Fore.RED+f"Error trying to play audio:\n{e}")
-        if elevenLabs == True: # If ElevenLabs is used, this error likely occured because of wrong API Key
-            print(Style.BRIGHT+Fore.RED+"WARNING: This error likely occured because the ElvenLabs API Key is wrong. Please provide an API Key in apiKeys.py to use ElevenLabs.")
+            voice = texttospeech.VoiceSelectionParams(
+                language_code="sv-SE",
+                name=googleTTS_name,
+                ssml_gender=ssml_gender,
+            )
+            audio_config = texttospeech.AudioConfig(
+                audio_encoding=texttospeech.AudioEncoding.MP3
+            )
+            response = client.synthesize_speech(
+                request={"input": input_text, "voice": voice, "audio_config": audio_config}
+            )
+            with open(textToSpeechFilePath, "wb") as out:
+                out.write(response.audio_content)
+            playAudio(language)
+        except Exception as e:
+            print(Style.BRIGHT+Fore.RED+f"Google TTS Error:\n{e}")
             print("Using offline text-to-speech instead...")
             offlineTextToSpeech(text)
-            elevenLabs = False
             return None
+        return textToSpeechFilePath
 
-    return textToSpeechFilePath
+    # English: try the selected provider first, then the other, then offline TTS
+    for name, generate in englishTTSproviders():
+        try:
+            generate(text) # Download/decode the audio to textToSpeechFilePath
+        except Exception as e: # The request to the provider failed
+            if isAuthError(e):
+                print(Style.BRIGHT+Fore.RED+f"{name} Error (likely a wrong/missing API Key in apiKeys.py):\n{e}")
+                disableTTSprovider(name)
+            else:
+                print(Style.BRIGHT+Fore.RED+f"{name} Error:\n{e}")
+            print("Trying the next text-to-speech option...")
+            continue
+        try:
+            playAudio(language)
+            return textToSpeechFilePath
+        except Exception as e: # The audio could not be played (corrupt audio or wrong API Key)
+            print(Style.BRIGHT+Fore.RED+f"Error trying to play {name} audio (the API Key may be wrong):\n{e}")
+            disableTTSprovider(name)
+            print("Trying the next text-to-speech option...")
+            continue
+
+    print("Using offline text-to-speech instead...")
+    offlineTextToSpeech(text)
+    return None
 
 def offlineTextToSpeech(text):
     tts.say(text)

apiKeys.py

@@ -7,6 +7,9 @@
 # ElevenLabs
 XI_API_KEY = "{KEY_HERE}"
 
+# 60db.ai (Text-To-Speech)
+SIXTYDB_API_KEY = "{KEY_HERE}"
+
 # Google Search Engine
 googleCustomSearchAPI = "{KEY_HERE}" # Google Custom Search API Key (developerKey)
 googleSearchEngineID = "{ENGINE_ID_HERE}" # Google Search Engine ID (cx)